Investigation into the Creation of an Algorithm Processor Module (APM), 10-R9581

Principal Investigators
Stephen R Johns
Christopher M Auyeung
Kevin T Chan
Dedaimia Kozlovsky

Inclusive Dates:  10/01/05 – 09/29/06

Background - Intelligent transportation systems are comprised of multiple components, including various types of roadway field equipment that provide data about traffic and environmental conditions. These components provide independent data and, to date, no method has been developed to aggregate those individual pieces of data together in real time for more advanced analysis. Significant benefits to transportation management operations could be realized if the data from these different subsystems could be combined, correlated, and analyzed. A broader, more advanced, and more thorough range of analysis could be performed on the aggregated data.

Approach - An incremental approach was used, with each task providing the information necessary to guide the next task. First, existing subsystems were inspected to gain a better understanding of the structure of the data they produce. The data were analyzed to determine the units of measure, the size of the data, the frequency at which the data is produced, the time periods for which the data is valid, the format in which the data is provided, and the areas of analysis to which the data are relevant. Next, methods for combining heterogeneous data streams were investigated. This task allowed the team to identify and develop approaches and techniques for merging raw data into Virtual Data Channels (VDCs).

Accomplishments - A mechanism, including an Extensible Markup Language (XML) schema, was developed to document and describe data sources. An XML schema was also developed to describe VDCs. Based on a review of current best practices, a generic framework for performing data clustering was developed. The framework is used by the Algorithm Processor Module (APM) to identify which pieces of data from the various data sources can be combined. The specific algorithms used to combine data in the VDCs are included in the XML configuration files. These algorithms are compiled and executed at run time. An application programming interface was implemented in C#, which allows applications to work with the APM. A prototype graphical user interface that provides users with an easy-to-use graphical programming language was created. Users simply drag-and-drop the data elements, operations, and actions they need into the program panel to create working programs. This project resulted in a prototype system that allows end users to quickly and easily create programs that make use of information extracted from aggregated, existing data sources, both static and real-time.

2006 Program Home